Text Mining and Sentiment Analysis of SuperBowl LII

Sentiment Analysis of Tweets from SuperBowl LII

Do you own a business and wonder what people aka customers are saying about your company?  Using text mining and sentiment analysis enables you to uncover the words your customers are using and the positive or negative sentiment of those words.  Text mining involves taking text from any source and breaking it down into the core words.   Sentiment analysis then applies the score to the words from -1 (the worst sentiment) to 0 (neutral sentiment) to 1 (highest positive sentiment).  The scores of all the words found create the overall sentiment for that encounter.

To show this in action, I ran sentiment analysis on Tweets the day after SuperBowl LII (2/5/2018).  In one example, someone tweeted “Congratulations Philadelphia SuperBowl champions”.  The word champion has a positive sentiment and has a positive sentiment score of  .70.  When you take all of the words in an example and score them you get the total sentiment for that tweet.  For example, someone could write “Eagle Qback was on fire it to Win the SBowl”.  In this example, the context is positive, however, the use of “fire” will bring the sentiment score down.  You can apply this to twitter, facebook or other social media sites.  In addition, if you apply this to your internal operations such as customer service notes you can find the sentiment of what you customers are telling your reps.

Here are some examples that show the distribution of sentiment for various topics.  I used RapidMiner and the Twitter connector to get 1000 Tweets on the word SuperBowl and the words halftime show.

Using the word SuperBowl, I found the following distribution of tweets.  The frequency of the score is on the y-axis and the sentiment score is on the x-axis.  This shows most tweets about 240, are neutral between -.05 and .05.  There were slightly more positive tweets from .05 to 2.0.  However, overall there were more tweets with negative sentiment in this sample.

Sentiment Analysis

For the next example, I wanted to see what people were saying about the halftime show.  Using Sentiment Analysis and Text Mining we can see that most people tweeting about the halftime show were more positive in their tweets than negative.  With over 325 of the 1000 scoring between .07 and .12.

Text Mining and Sentiment Analysis

Applying Sentiment Analysis to Your Business

As you can see from the examples above.  I was able to quickly query Twitter and determine what the customers of the SuperBowl were spreading on social media.  In an organization, you can apply the same techniques to monitor what your customers are saying about your brand.  In addition, in any applications where you store text data such as claims or customer support.  You can mine the text to find the common words that are being entered into the system.  Then apply sentiment analysis to see what your team is recording from your customers.


Learn more about Sentiment Analysis